AI Pipeline for Decision-Ready ITSM Intelligence
A recent study released on arXiv introduces a novel approach to enhance the utilization of IT service management (ITSM) data within organizations. By leveraging Large Language Models (LLMs), the research focuses on transforming raw ticket data into practical insights for sales teams and executives. Utilizing clustering techniques, the system organizes ticket information into main and sub-topics effectively. In an evaluation involving six artifacts assessed by five reviewers from Sales Engineering and customer support, key metrics such as interpretability and actionability scored over 4.0 out of 5.0, particularly excelling in trust, emphasizing the significance of effective ITSM analytics.
Key facts
- Paper arXiv:2608.12670 presents an AI pipeline for ITSM intelligence.
- Pipeline uses LLM-based schema normalization, HDBSCAN clustering, and hierarchical agglomerative clustering.
- Evaluation involved six artifacts and five raters from Sales Engineering and customer success.
- All four decision-support metrics averaged above 4.0 out of 5.0.
- Trust was the most consistent signal among the metrics.
- The pipeline transforms raw ITSM exports into multilevel decision-support artifacts.
- The study follows design science research principles.
- ITSM analytics is framed as an Information Systems (IS) problem.
Entities
Institutions
- arXiv